Seaborn barplot with x_order: AttributeError: 'bool' object has no attribute 'sum'

Question

I am trying to plot simple barcharts with seaborn.barplot(). In the simplest case it works well, x can by a vector of strings, or numbers:

import numpy as np
import seaborn as sns
import matplotplib.pyplot as plt
fig, ax = plt.subplots()
ax = sns.barplot(np.array(['a','b','c']), y = np.array([1,2,3]))
fig.tight_layout()
fig.savefig('test.pdf')

To order the bars in a custom way, there is the x_order argument, which is a list-like object with indexes. If x itself is numeric, it also works well:

x = np.array([2, 0, 1])
y = np.array([3, 4, 2])
sns.barplot(x, y = y, x_order = list(x.argsort()))

However, if x is not numeric, it gives an error, even if I try to order by other numeric vector, or by ordering the string vector itself:

x = np.array(['b', 'c', 'a'])
y = np.array([3, 4, 2])
sns.barplot(x, y = y, x_order = list(x.argsort()))
sns.barplot(x, y = y, x_order = [2, 0, 1])

Then I get the error AttributeError: 'bool' object has no attribute 'sum'. I haven't found much about this, and I am wondering how to do this simple ordering properly.

Answer 1

x_order should be a list of the labels, not the list of indexes. In other words, in the latter case you would just want x_order=['a', 'b', 'c'].